Almost rolling motion: An investigation of rolling grooved cylinders

Almost rolling motion: An investigation of rolling grooved cylinders

Lawrence R. Mead and Frank W. Bentrem
Department of Physics and Astronomy, University of Southern Mississippi, Hattiesburg, Mississippi 39406-5046
(Dated: Received 21 July 1997; accepted 12 September 1997)

We examine the dynamics of cylinders that are grooved to form teeth for rolling motion down
an inclined plane. The grooved cylinders are experimentally found to reach a terminal velocity.
This result can be explained by the inclusion of inelastic processes which occur whenever a tooth
hits the surface. The fraction of the angular velocity that is lost during an inelastic collision is
phenomenologically found to be proportional to 2sin^ tv/N — asin'^ t^/N, and the method of least
squares is used to find the constant a = 0.98. The adjusted theoretical results for the time of rolling
as well as for terminal velocity are found to be in good agreement with the experimental results,
(c) 1998 American Association of Physics Teachers. This article may be downloaded for personal use only.
Any other use requires prior permission of the author and the American Association of Physics Teachers.

X

I. INTRODUCTION

A few years ago an article appeared in this journal^
which described how the fractal dimension of a randomly
crumpled surface might be determined by roUing it down
an inclined plane and measuring its moment of inertia
/. In this paper, the claim is made that the moment of
inertia of a hypersphere is given by

(1)

where 7 is a constant that depends on the dimension of
the hypersphere, M is its mass, and R is its radius. This
conjecture was used to attempt to measure the fractal
dimension of tightly crumpled aluminum foil balls. How-
ever, it has been shown^ that this same sort of variation
in the dynamically measured moment of inertia occurs
with symmetrically grooved cylinders, such as the one
depicted in Fig. [I] which are not fractals.

These observations prompt one to ask to what extent
the "almost rolling" motion of the grooved cylinders can
be analyzed and understood. In particular, can the mo-
tion of regular grooved cylinders be accurately described
in any simple way by dynamics and conservation laws
of basic physics? Furthermore, there does not seem to
be any reference to this kind of "nearly rolling" motion
in the physics literature. Advanced engineering texts?.!^
discuss rolling motion and contact, collisions, enmeshed
gears, etc. However, the "rolling" of a "gear" or "slot-
ted cylinder" does not seem to have been studied. This
"rolling" motion of a grooved cylinder is similar to the
motion of a smooth cylinder rolling on a rough surface. A
rough surface tends to slow down a rolling smooth cylin-
der and is considered part of rolling friction,* although
the dynamics of this effect has not been well analyzed.

In this paper, we report a study of the "nearly rolling"
motion of regularly grooved cylinders down an incline of
fixed angle. In Sec. II, the motion of such objects is an-
alyzed using Lagrangian dynamics and making the sim-
plifying assumption that no energy is dissipated by the
"collisions" between the teeth and the inclined plane, but
that the total energy of the cylinder is completely con-
served. A recursive scheme is then derived for predicting

how long a cylinder will take to roll a given, but vari-
able, distance along an incline. Computer calculations
yield predictions for these rolling times. In Sec. HI, ex-
periments are described which were used to measure the
rolling times and to compare with the theoretical pre-
dictions. The results of the first of these measurements
are in marked disagreement with the theoretical predic-
tions, with the assumption that the "collisions" are com-
pletely elastic. Indeed, the experimental data show that
the cylinders reach terminal velocity, in contrast to the
theoretical predictions.

A more precise theoretical model of this "near rolling"
motion is then presented in Sec. IV, which allows for
inelastic processes to occur as the cylinder "rolls." Pre-
dictions of this modified theory for the measured rolling
times and terminal velocities are remarkably accurate.

II. DYNAMICS

Our problem consists of calculating the time it takes
for a grooved cylinder to rotate about one tooth as it is
rolling down an inclined plane. This section contains a
discussion of the dynamics for this motion and a deriva-
tion of an expression for the time of rolling.

A. Rotational kinetic energy

We will make the assumption that the grooved cylinder
is rolling without slipping, i.e., the single pivot point, P in
Fig. 2, does not slide. So as shown in Fig. 2, the cylinder
will simply pivot about each tooth at the point P until
the next tooth comes in contact with the inclined plane.
In Fig. 2, -0 is the elevation angle of the incline and 9 is
the angle through which the center of mass has rotated
starting from a vertical position. The kinetic energy T of
the rolling grooved cylinder is entirely rotational (about
P) and is given by

T

1

(2)

2

FIG. 1: A grooved cylinder with N = 6 teeth.

FIG. 2: A grooved cylinder with N = 6 teeth.

where Ip is the moment of inertia of the grooved cyhnder
about the edge of the tooth in contact with the inchned
plane (point P), and 6 is the angular velocity about P.

B. Time of rolling of the grooved cylinder

Using the parallel axis theorem, Eq. (2) becomes
1

(3)

where M is the mass of the grooved cylinder, R is the
outer radius (i.e., the distance from the center of mass

to the edge of a tooth) . and /cm is its moment of inertia
about an axis through the center of mass. If we take

the potential energy to be zero when is zero, then the
potential energy U can be expressed as

U = Mgh = MgR{cos0 - 1),

(4)

using the fact that the change in height h of the center
of mass as a function oi 9 is h = R{cos9 — 1). So the
Lagrangian L is

L = i (Mi?2 + J,^) + MgR{l - cos 0). (5)

For our problem Lagrange's equation is

d_dL _dL_^
di~d0 ~ '80 ~ '

We can solve Lagrange's equation to obtain the result

•• MgRsm9

(6)

Mi?2 + 7c„'

(7)

where 9 is the angular acceleration about point P. Mul-
tiplying both sides of Eq. (7) by ^ and integrating over
time leads to

C-^cos6i,

(8)

where C is the integration constant and ^ is defined as
2MgR/{MR'^ + /cm). This is a separable differential
equation and can be solved to find the time t it takes
for the grooved cylinder to rotate about a given tooth,
starting at the instant the previous tooth leaves the sur-
face of the inclined plane and ending with the instant the
next tooth strikes the surface. The result is

t

d9

s/C-^cos9'

(9)

where 0o is the initial value for the angle and ^/ is the
final angle. This is an elliptic integral and must be eval-
uated numerically. We find by geometry that the initial
and final angles are ip + w/N and tjj — n/N, respectively,
where tjj is the elevation angle of the inclined plane (see
Fig. 3). So we have

d0

^-tt/at \/C - C cos (

(10)

To solve for the integration constant, evaluate Eq. (8) at
t = Q and rearrange it to give

C = (Jo + ^ cos - ^) ,

(11)

where we have used the initial condition that 0^] is the
initial angular velocity wq when = ^ — tt/N. Similarly,
we know that 9 is the final angular velocity u)f when
9 = tp + n/N, so that Eq. (8) becomes

u;} = C = icos (V' + ^)-

(12)

3

FIG. 3: The initial and final angles for rotation about a tooth.

Since for the moment we assume no loss of kinetic en-
ergy from the teeth striking the inchnc, the final angular
velocity for the rotation about one tooth is the initial
angular velocity for the rotation about the next tooth.
Substitute Eq. (12) into Eq. (11) and rearrange to get
the recursion relation

Cfc = Ck-i + ^

cos

TT \

+ n)

(13)

where Ck is the integration constant for the fcth rotation
about a tooth and Ck-i is the integration constant for the
k — 1th rotation. We can determine Ci by substituting
into Eq. (11) the value of the initial angular velocity
for t = 0.

The total time r it takes for the grooved cylinder to roll

a given distance is just the sum over the time integrals
from Eq. (10) for each rotation about a tooth. We write

(14)

fe=i

where n is the total number of rotations and tk is the time
integral for the fcth rotation. It can be seen by geometry
that

n =

2Rsm{-K/N)

(15)

where L is the distance traveled along the inclined plane.

We need n to be an integer so the greatest integer func-
tion is used to eliminate any partial rotation at the end
of the track. So we are left with the computation of

■^{J^P-n/N VCfc - ^ COS 6*

for the total time, where Ck is given by Eq. (13).

(16)

C. Moment of inertia

The moment of inertia about the center of mass 7cm is
calculated by breaking up the grooved cylinder into tri-
angles as shown in Fig. 4 and performing the integration
over each triangle and summing, so that

Ic

^ Ip dx

dy (x

y

(17)

all A's

where I is the length of the cylinder and p is the volume
density. The volume density is defined as the mass per
unit volume, so we find that

M

P =

(18)

lNrRsm{iT/N)'

where r is the distance from the center of mass to the
inside cut of the grooves (see Fig. 1). So Eq. (17) becomes

x-^ M

^ NrR sm(Tr/N)

all A's \ I )

j dx j dy{x''+y^). (19)

The result for the moment of inertia is

Mr^ cos(7r/iV) / 2tt
/cm= 2 + cos-

2M

r2 sin2(7r/iV)

R{R-rcos{Tr/N)) i 3(i? - r cos(7r/iV))2
E3(i? = .cos^)-^(i?^-r^cos^^)

R{R^ -r"^ cos^

N

+

R

R^

r^cos^-^)

4 4

r cos

4 4

r cos

N

TT
N

(20)

That this result for /cm yields the correct value of \MR?
for a smooth cylinder can be verified by taking iV — > oo
and then setting r = R. Notice that when we substitute
Eq. (20) into Eq. (16) the mass cancels out [substitute
Eq. (20) into So the time of rolling is independent of
the mass of the grooved cylinder just as it is for a perfect
cylinder.

D. Computer model

A computer program was written to numerically eval-
uate the integrals in Eq. (16). The first term in the sum
in Eq. (16) contains Ci , which can be calculated from
Eq. (11) for an initial angular velocity. If the grooved
cylinder is started from rest with two teeth in contact
with the surface, then there is some critical angle of el-
evation ipc below which the cylinder will not roll. This
critical angle depends on the number of teeth and is given

by

TT

TV'

(21)

4

FIG. 4: A grooved cylinder divided into triangles for the pur-
pose of calculating the moment of inertia.

To ensure that all the grooved cylinders will be able to
roll, we let the initial angular velocity be the angular ve-
locity that the cylinder would have if it started from rest
at a balance point 9 = i). This corresponds to balancing
a grooved cylinder and starting the timer when the first
tooth strikes the surface of the inclined plane. We use
conservation of energy

AT = -AU (22)

to find the angular velocity, where AT is the change in
kinetic energy and AU is the change in potential energy.

We have

i (Mi?2 + /,„) _ ^ MgR{cos9i - cos Of) ,

(23)

where 9i and 9 / are the initial and final angular velocities,
and 9i and 9f are the initial and final angles, respectively.
The final angle and final angular velocity correspond to
the instant the timer is started. So we can use the final
angular velocity as the initial angular velocity for the
rotation about the first tooth. Now Eq. (11) is used to
find the first integration constant Ci. All of the other
integration constants can be found by using the recursion
relation given by Eq. (13). We now have everything that
we need in order to compute the time of rolling for a
given grooved cylinder rolling a prescribed distance down
an inclined plane. The computer model can be easily
verified with a smooth cylinder simulation, i.e., take a
large value for N and take r R. (We cannot have
r = R because this would lead to division by zero in the
calculation of the moment of inertia). If we let a smooth
cylinder roll a distance of 27r cm at an elevation angle of
30°, we find the analytical value for the time of rolling
to be 0.196 134 s. For the computer calculation we use
the values i? = 1 cm, r = 0.999 999 99 cm, TV = 100000

TABLE I: Intrinsic data for the grooved cylinders.

N

M{g)

R (cm)

r (cm)

/cm ( g cm^)

12

9.726

1.000

0.575

3.06

18

12.860

1.000

0.735

4.85

30

14.090

1.000

0.845

6.00

45

14.090

1.000

0.900

6.72

60

15.425

1.000

0.945

7.29

teeth, and L = 2t:R = 6.283 185 cm. The numerically
computed value for the time of rolling is 0.196 123 s,
which is accurate to five digits.

III. EXPERIMENTAL PROCEDURE

Several grooved cylinders with various numbers of
teeth were needed in order to find out how the number of
teeth affects the rolling motion. We used five solid alu-
minum cylinders which had been cut to have 12, 18, 30,
45, and 60 teeth. Table I records the mass, inside and
outside radii, and moment of inertia for these grooved
cylinders. The moment of inertia was calculated using
Eq. (20). The grooved cylinders were 2.000 cm long and
had an outside radius R of 1.000 cm. The rolling surface
used was a sheet of Plexiglas, which was rigidly attached
to a thick flat board. The elevation angle for this exper-
iment was set a.t ip = 1.39 ± 0.02°. An elevation angle
that is too small (i.e., less than 1°) prevents the 12-tooth
cylinder from proceeding down the inclined plane when
released from rest. On the other hand, if an angle greater
than about 3° is used, the support points for the cylin-
ders can actually leave the surface due to bouncing.

In order to measure the time of rolling a model ME-
9215A Pasco Scientific Photogate Timer with Memory
was attached to the board with one photogate positioned
at the point where the cylinders would start to roll and
another photogate whose position could be adjusted in
10-cm increments to a given distance from the first pho-
togate. The exact distance between the photogate beams
can be obtained by sliding a strip of cardboard along
the incline until it "trips" the first photogate and mark-
ing this position. Sliding the cardboard further until it
"trips" the second photogate and marking this position
allows one to measure the distance between the marks,
which is the distance between the photogate beams.

It was observed that the rolling grooved cylinders
would eventually reach terminal velocity. A microphone
was plugged into a model 54600A 100-MHz, two-channel
HewlettPackard digital oscilloscope with memory in or-
der to pick up the clicking sound of the teeth of the rolling
grooved cylinders as they hit the Plexiglas surface. Each
click of a tooth created a spike on the oscilloscope. At
terminal velocity the clicking sound had a constant fre-
quency, which yielded a constant spacing between spikes
on the oscilloscope. The grooved cylinders were rolled a

5

TABLE II: Time for rolling 80 cm (energy conservation
model) .

No. Teeth

Theoretical time (s)

Measured time (s)

12

2.74 ± 0.12

10.40 ± 0.07

18

2.86 ±0.08

7.29 ± 0.05

30

2.95 ±0.08

5.70 ± 0.03

45

3.01 ± 0.06

5.00 ± 0.03

60

3.05 ± 0.06

4.57 ±0.08

given distance at which point the screen on the oscillo-
scope was frozen. Two vertical cursors were positioned
on the peaks of two distinct spikes. The oscilloscope in-
dicated the period between the two cursors. The period
was divided by the distance between the teeth to get the
velocity. This method for measuring the velocity of the
rolling cylinders worked quite well for the 12-, 18-, and
30-toothed cylinders. However, when this technique was
employed for the 45- and 60-toothed cylinders, it was
difficult to separate the impact signal from the noise and
echoes. In order to measure the terminal velocity for
these cylinders two photogates were placed about 10-cm
apart at the place the velocity was to be measured. The
photogates would measure the time it took for the cylin-
ders to roll from one gate to the other. The distance
between the photogates divided by the time was the ve-
locity. That the cylinders reached terminal velocity was
verified by measuring the velocities at different rolling
distances.

In the first part of the experiment the time was mea-
sured for the grooved cylinders to roll 80.02 cm. Each
cylinder was given five trials. The difficulty in this ex-
periment was in rolling the cylinders through the narrow
photogates (6.6 cm wide). These times are compared
in Table II with the times predicted from the computer
program based on the Lagrangian mechanics. As one can
plainly see, there is little agreement between the theoret-
ical and experimental values. The time values are clos-
est for the 60-toothed cylinder. As the number of teeth
is decreased, the theoretical values slowly decrease but
the experimental values increase significantly. In fact,
the cylinders were observed to reach a terminal velocity
which was lower not higher -for fewer teeth.

The standard deviation is given as the estimated error
for the measured times. Errors in the theoretical times
are due to the uncertainty in the elevation angle and the
final rotation about the last tooth and will be discussed
later.

the loss of kinetic energy caused by the teeth striking
the surface of the inclined plane is negligible. The first
assumption is justified by the small angle of elevation;
therefore, the loss of kinetic energy caused by the inelas-
tic collisions apparently cannot be neglected.

A. Angular momentum model

In retrospect, that the collisions are inelastic should
not be surprising since each tooth strikes the surface and
then does not rebound but stays at that point to act
as a pivot. The duration of the collision extends from
the moment a tooth strikes the surface to the moment
the previous tooth lifts off the surface. Conservation of
angular momentum can be used to find out how much
energy is lost in these collisions; for now we will assume
an instantaneous collision. Choose a point on the incline
where the front tooth is about to hit (just before the col-
lision) as the origin. In general, the angular momentum
L of the cylinder about that point is

L = mr X Vcm + /cmW,

(24)

where m is the mass of the cylinder, r is the position of
the center of mass relative to that origin, and u) is the
angular velocity about the center of mass. We note with
the aid of Fig. 5 that the initial velocity vector, which
has the magnitude Vi = ROi, makes angle 7r/2 — 27r/7V
with Tj. Thus the initial angular momentum just before
the collision is

Li

,'2n

MR cos ( —

+ /c>

(25)

Similarly, the angle between v/ and r is 90° just after
the collision; hence.

Lf = {MR^+I,^) Of.

(26)

In either case, v = R9. By conservation of angular mo-
mentum we equate Eqs. (25) and (26) and rearrange so
that

2 2 TT

sm —

+ X N

(27)

where x is defined as /cm/ {MR^^. The initial angular

velocity 9i just before the collision is the final angular
velocity 6f after the partial rotation given by Eq. (12).
So upon substitution we get

IV. A MORE COMPLETE THEORETICAL
MODEL

In constructing the theoretical model of Sec. II, two

assumptions were used: that the cylinders are rotating
about their point of support without slipping and that

sm —

1 + X N

C — ^ cos

(^ + ^)], (28)

where ^ has been defined previously to be
2MgR/ (MR"^ + Icm)- Since the final angular ve-
locity 6f for a rotation about one tooth is the initial

6

TABLE III: Time for rolling 80 cm (angular momentum
model) .

No. Teeth

Theoretical time (s)

Measured time (s)

12

11.58 ±0.19

10.40 ± 0.07

18

7.72 ±0.10

7.29 ± 0.05

30

5.95 ±0.09

5.70 ± 0.03

45

5.09 ± 0.07

5.00 ± 0.03

60

4.64 ± 0.07

4.57 ± 0.08

angular velocity for the rotation about the next tooth,
Eq. (11) becomes

Cfe

sin —

1 + X N

Ck-i - ^cos

(29)

Using this recursion relation in the computer program to
find the integration constant yields the results given in
Table III.

B. Model with finite-time collisions

These results are somewhat better than the previous
calculations. However, since we really do not have in-
stantaneous collisions some angular momentum must be
added back to the cylinder after each collision due to a net
external torque which acts over a small but finite period
of time. During this time the surface of the inclined plane
is slightly deformed at the point of contact for both teeth,
as depicted in Fig. 6. The teeth themselves can also be
deformed. Figure 6 shows the relevant forces that cause
the net torque about the center of mass. The forces la-
beled A^i and N2 are normal forces, and those labeled /i

and /2 are due to friction. Without knowing the detailed
dynamics of the collision it is natural to try to find a
phenomenological term that adds the correct amount to
the angular momentum.

If an additive term can be found which fits the mea-
surements for 80-cm rolling times, then using this addi-
tive term we can compare the theoretical and measured
results for the time of rolling at different lengths and ter-
minal velocity. The 80-cm rolling times were chosen for
the fit because of the relatively low errors in the mea-
sured times. A good fit was obtained using a term that
is proportional to sin^(7r/A^). The proportionality con-
stant a was found by using the method of least squares
to be a = 0.98 ± 0.05. Equation (27) becomes

Of

1

■ sin

+

1+X N 1+x

■ sm

TT

N

(30)

This leads to the new recursion relation for the integra-
tion constant

1 -

sm — + sm —

1+X ^ 1+X N

Cfe_i-^cos(v + ^) +Ccos(V-^)

(31)

Table IV compares the theoretical and measured rolling
times for 80 cm using this additional angular momentum
term.

An accurate description of the "rolling" motion of
grooved cylinders, in addition to predicting rolling times,
should be able to predict the terminal velocities of the
cylinders. Experiment showed that each of our grooved
cylinders had reached terminal velocity after rolling 60
cm. The velocities of the cylinders were measured af-
ter rolling 115 cm and, taking an average over ten trials,
were compared to the predictions from our modified the-
ory. Table V lists these values.

7

TABLE IV: Time for rolling 80 cm (model with finite-time
collisions) .

No. Teeth Theoretical time (s)

12
18
30
45
60

10.35 ±0.19
7.40 ±0.12
5.84 ±0.11
5.04 ± 0.09
4.61 ± 0.08

Measured time (s)

10.40 ± 0.07
7.29 ± 0.05
5.70 ±0.03
5.00 ±0.03
4.57 ±0.08

TABLE VII: Theoretical rolling times at various lengths.
No. Teeth 20 cm (s) 40 cm (s) 60 cm (s) 80 cm (s)

12 2.77 ±0.12 5.32 ±0.13

18 2.14 ±0.10 3.91 ±0.10

30 1.88 ±0.10 3.21 ±0.10

45 1.78 ±0.08 2.89 ±0.09

60 1.73 ±0.07 2.73 ±0.07

7.80 ±0.15 10.35 ±0.17

5.64 ±0.11 7.40 ±0.12

4.53 ±0.10 5.84 ±0.09

3.97 ±0.09 5.04 ±0.07

3.68 ±0.07 4.61 ±0.07

TABLE V: Terminal velocity.

No. Teeth

Theoretical velocity

(cm/s)

Measured velocity
(cm/s)

12

7.92 ±0.10

8.08 ±0.26

18

11.43 ± 0.10

11.78 ± 0.42

30

15.23 ±0.12

15.25 ±0.10

45

18.67 ±0.14

18.46 ±0.15

60

21.52 ±0.16

21.57 ±0.24

Rolling times were measured for several different
lengths. The results are given in Table VI. Table VII
contains the predicted rolling times for the lengths used
in Table VI. These predictions come from the modified
theory which includes inelastic processes. The propor-
tionality constant a = 0.98 comes from the least-squares
best fit to the 80-cm rolling times. The agreement be-
tween the measured and predicted results is remarkable
considering we used only a single best-fit parameter. The
only measured value which did not fall within the error
estimates of the predicted value is the 40-cm rolling time
for the 12-toothed cylinder; the most difficult case from
which to obtain reliable results.

V. CONCLUSIONS

We found that the "rolling" motion of a symmetrically
grooved cylinder can be accurately described by assuming
that the cylinder rotates about each tooth without slip-
ping. Angular momentum is nearly conserved during the
collision, however, inelastic processes cause a net torque
to act on the cylinder over a small but finite period of
collision time. (Recall that the collision period extends

TABLE VI: Measured rolling times at various lengths.
No. Teeth 20 cm (s) 40 cm (s) 60 cm (s) 80 cm (s)

12 2.98 ±0.09 5.60 ±0.05 ~7? 10.40 ± 0.07

18 2.19 ± 0.05 3.93 ±0.05 ...^ 7.29 ± 0.05

30 1.95 ± 0.05 3.19 ±0.02 4.53 ± 0.06 5.70 ±0.03

45 1.82 ±0.08 2.89 ±0.05 4.02 ± 0.06 5.00 ± 0.03

60 1.76 ± 0.04 2.75 ±0.04 3.75 ±0.11 4.57 ± 0.08

from the moment the tooth comes in contact with the
inclined plane to the moment the previous tooth lifts off
the plane.)

The time it takes a grooved cylinder to "roll" a distance
L along an incline with an elevation angle i]} is given by
Eqs. (15) and (16). The integration constant in Eq. (16)
is given by Eq. (31), where the proportionality constant
a was found to be 0.98 ± 0.05 using a least-squares best
fit to the 80-cm rolling times.

This constant probably depends on the coefficient of
restitution of the cylinders on the incline. The constant a
may also depend on the elevation angle, mass and radii of
the grooved cylinders, and elasticity (leading to bending
moments) of the cylinders and the inclined plane.

The estimated errors given for the measured rolling
times and terminal velocities are the standard deviations
of the measurements. The estimated theoretical errors
come from two sources. The first error source is the un-
certainty in the elevation angle. The angle was measured
to an accuracy within 0.02°. The amount of error from
this source was estimated by using this variation in the
computer program (1.37° instead of 1.39°) and finding
the difference from the originally computed value. An-
other significant error source comes from not having a
complete final rotation about the last tooth. The cylin-
der "trips" the final timing gate before the final rotation
is complete. The predicted times include half of the time
it takes to make the final rotation which is also the esti-
mated error from this source.

Additional error comes from the cylinder's tendency
to travel in a slightly curved path. This tendency would
clearly result in measured times which are longer than
for the assumed straight line path. One final source of
error comes from the uncertainty in the initial angle and
the initial angular velocity of the grooved cylinders. The
cylinders were released from a vertical balanced position
and "tripped" the initial timing gate just after release.
It was assumed that the timer started after the cylinder
had rotated 0.5°. The amount of error from the latter two
sources is not likely to be large and would be difficult to
estimate.

Future study on "rolling" motion of grooved cylinders
could focus on determining how the proportionality con-
stant a depends on the various parameters mentioned
above. Grooved cylinders that are not symmetrical would
also be interesting to study. Finally, spheres that are
"grooved" in three dimensions such as tightly crumpled

8

aluminum foil balls could be investigated. This could
contribute to a better understanding of rolling friction
caused by surface roughness.

Acknowledgments

Special thanks go to Dr. Raymond Folse and
Glen Messer for their valuable insight and suggestions

throughout the experimental phase. Cooperation from
the School of Engineering Technology of the University of
Southern Mississippi in providing a digital oscilloscope is
much appreciated. Finally, we would like to express grat-
itude for the funds received from the College of Science
and Technology (USM) for manufacturing the grooved
cylinders.

^ F. F. Lima, V. M. Olivcira, and M. A. F. Gomes, "A
Galilean experiment to measure a fractal dimension," Am.
J. Phys. 61, 421-422 (1993).

^ L. R. Mead, R. F. Folse, and A. Cole, "Rolling motion and
crumpled surfaces," Am. J. Phys. 63, 746-749 (1995).

^ K. L. Johnson, Contact Mechanics (Cambridge U.P., Cam-
bridge, 1985), pp. 242-311.

" C. E. Wilson, J. P. Sadler, and W. J. Michels, Kinematics

and Dynamics of Machinery (Harper & Row, New York,
1983), pp. 355-385.
® No data are available for 60-cm rolling times for the cylin-
ders with 12 and 18 teeth because of the problem of rolling
these cylinders through the narrow timing gates. The 80-cm
rolling times were obtained with great difficulty.